Some files to download:
X-Org.conf
GRUB menu.lst with options
X-Org keyboard Layout to get Umlauts from the US Keys
After some promising reports, I updated to Debian unstable and Kernel 2.6.15. This solves several issues. For example, it is no possible to get the clock to normal spee with the option 'diable_timer_pin_1' which seems to have no undesired side effects (so ACPI still running and so on). Additionally with the 'ec_bust=1' option, you get one step closer to the battery, where there is actually acknowledged by the system that there is a battery present, along ith some static data. However, it still seems impossible to get the battery state without DSDT patching.
Best thing of all, finally X-Org and getting rid of the **** commercial ATI driver. The default X-Org driver will perform brightly as expected, even giving you a Clone mode. Be aware, however, that you run into the same problems as with the binary driver, so initially, the picture will be sent to the CRT or DVI out but not to the internal flat panel. Same hack as with the commercial driver.
After the update everything worked fine up to the moment, where I wanted to use a PCMCIA/CF adaptor to access the pictures on my trusted camera. It did not work. I got the message "hde: hde 1" repeated over and over again, obviously some sort of bug.
To ease the tention with the interested reader, udev is responsible for this and, if google can be believed, udev and pcmcia in general don't like each other. The solution here is to do a step back from udev to hotplug. Which is kinda hairy. The initial ramdisk is different, depending, whether you use hotplug or udev. So, using dselect, to uninstall udev, it will tell you to uninstall the kernel also, because of a dependency problem. Not nice. You need another mkinitrd program, to be able to keep the kernel, however, that does not change the wrong (w.r.t. hotplug) initial ramdisk image already installed. You have to make a new initial ramdisk image and before you do, you have to add "atiixp" to the /etc/mkinitrd/modules file or you won't have dma running! Afterwards execute `mkinitrd -o /boot/name_of_initrd_image' and to be safe, you should backup your old initial ramdisk beforehand.
By the way, it seems, if you are using the commercial ati driver for the chipset, you are required to add three lines to the /etc/mkinitrd/modules: `ide-generic' `ata_piix' and `sd_mod'. However, I didn't try it myself.
For the interested, the problem here is, that the generic ide drivers are loaded, before the specialised drivers kick in (so generic ide is loaded within initrd, specialised afterwards, because hotplug seems less sophisticated here). That way, hard disk access is not controlled by the dma capable ati driver but only by the non-capable generic driver. This problem is quite common and in earlier times could be fixed by adding the driver modules like atiixp or via82cxxx to /etc/modules - initrd has changed that, so that /etc/modules is evaluated at a too late time point.
with a few minutes to spare, I got bluetooth, the touchpad and firewire running. I also got a clue about the battery and will pursue this hint this weekend
new wlan section down under
Remarkable thing is, once you set up such a page, you immediatly get help by other people and additional questions. I'll post the first category here and try to answer the second category.
Francois Wautier sent me a very helpful clue regarding the fast running clock. It is indeed a APIC problem and he suggested to use the boot option noapictimer. For him this does the trick on a somewhat similar notebook - for me it doesn't. Difference is, he is using kernel 2.6.12 while I'm stuck with 2.6.11 because of the ati binary driver (no driver for 2.6.12 yet). I've experimented with kernel boot options and it seems to me, noapictimer with kernel 2.6.11 does the same as noapic. noapic on my notebook gets the startup process to freeze when dealing with irq issues. Disabeling acpi too solves this issue, however, I decided to temporarily stick with the double time clock and still use acpi.
I wrote something about how not to select autodetec in the fglrxconfig to get the internal tft to work. When you chose 'single head' mode out of fglrxconfig, you won't be asked and the driver will assume you want it to use autoconfig (and therefore, you get no image). Try to use the 'clone' mode, even if you do not intend to connect an external monitor.
Initially, I questioned, whether I was too stupid or it wouldn't be supported. This question has been answered: Plug the firewire plug into the port with some force and everythin is running. I was surprised, because on my old notebook the plug went into the port with much less force. However, once done it works with the unpatched kernel drivers of 2.6.11.12. I have tested it only with an external Harddisk, maybe I will try ieee ethernet sometime in the future.
Finally, I did an update to debian unstable. Reason for this is the failure of my domain provider to react to support mails and get his imap server configured right, so I only get my mails with kde3.5 or Thunderbird *ranting off*
Finally I got the fast running clock under control without disabling USB and all the rest because the noapictimer seems to render acpi/apic more or less useless. The new option in 2.6.13 and above is "disable_timer_pin_1" which does the job nicely. Mouse is now running together with a slow clock and it does power off by itself. Whatelse can one ask?
I also got further with the battery issue, also from 2.6.13 on there is an option "ec_bust=1" to be added on bootup. This is close - finally I get an /proc/acpi/battery entry, telling me, the battery is present, if it is. Static data is being monitored, so last full capacity, design capacity and so on is available, current charge state isn't. Pity, but still, we're getting closer (by the way, I tried some acpi patches I found and none of them did the trick. Perhaps now?
I also got X working without the darn binary driver, as unstable comes with X.org. This does support the X700 nicely. Same problem as with the binary driver, though. Initially, X tries to communicate to the world via CRT, not LVDS, so you get a blank screen initially. If this happens to you, remember, its not frozen, so kill the X Server via ctrl-alt-backspace and shut down with ctrl-alt-del afterwards.
Rumours have it, it is possible to connect an external beamer on the running notebook with the X.org driver. I didn't (yet) manage to do that but will look into that. Any helpful hints in that directions will be violently caught, restrained, tested and then publicly taken apart.
Brandnew within this report, some of my config files: The X.org config The Grub menu.lst US_DE, a file, shamelessly stolen somewhere on the web, enabling me to use Umlauts on an English keyboard layout.
With the update, I have temporarily neither Wlan nor bluetooth tested, not yet time for that. Updates, as we get them
So I've spent some time now putting Debian stable on to the Ferrari 4000 Notebook made by Acer. This page has to be seen as a preliminary report, as I've not yet sorted all problems out, nor have I given up to solve some of them. So any suggestions on your side, questions or remarks are very welcome by mail to me
| processor | turion64 ML 37 (that is, 2000MHz, 1024 kb cache, 35W max power consumption) would be a 3200+ Athlon64 if it wasn't that economical ;-) |
| display | 15.4" 1680x1050 widescreen display |
| harddisk | 100 GB 5400 rpm |
| optical disk | dual layer DVD burner |
| communication | 10/100/1000 broadcom netxtreme broadcom wlan 802.11g some bluetooth |
| graphics adapter | Ati X700, 128 MB |
| memory | 2*512 MB DDR 333 memory mine is infineon, I've read of someone getting hynix... |
| chipset | Ati X200 |
| the rest | 1 PCCard Socket 4* USB connectors 1* Firewire 4 pin connector 1* D-Sub 15 VGA 1* DVI (yep, has both) 1* softmodem 1* composite TV-out 1* propetarian docing connector 1* bluetooth mouse 1* 4-in-1 card reader |
I bought the Ferrari at first, because it seems to be the only Athlon/Turion notebook around with a decent screen resolution. Maybe that's different, were you live but in Germany there seems an interesting correlation between Centrino and everything above XGA/WXGA. I bought it against the rule never to buy a notebook were you don't know anything about getting Linux running on that one, because I needed it really urgently and on the other hand, the software, I'll be using mainly on this notebook scales very well with the AMD architecture. I overruled my concerns, because Acer has made many notebooks running quite smoothly with Linux.
The design of the notebook is really cool. In contrast to the earlier Ferrari 3400 the ferrari red is less offensive in sight. All people I've asked so far admit, that Acer did an excellent design job on that one including the carbon lid. By the way, don't even think about removing the Ferrari stickers on the lid and on the inside. They are both embedded into the casing and everything done to remove them will make it looking worse.
Right, I really hate W-Lan. Yep, I really do. I consider any means of a computer connecting itself without me knowing to something dangerous. Everywhere I work I can get wired connection, so, why use something slower, less secure which might or might not be harmful? Apart from the hardcore solution, removing the W-Lan module, Acer allows for a softer way. There are two press-buttons in front of the notebook with LEDs behind them. Press the left button once to toggle the state of bluetooth, the right button to toggle WLan. Once the corresponding device is activated the buttons will be lit from behind. Bluetooth of course with a blue light, WLan with an orange one. I really like the idea
| device | status |
| display | working |
| tp-network | working |
| wifi-network | working |
| usb | working |
| powernow | working |
| sound | working |
| dma | working with later 2.6 kernels |
| ACPI | partially working |
| bluetooth | working |
| firewire | working |
| 4-in-1 cardreader | not tried -> reports indicate not working |
| touchpad | working |
In principle most things work out of the box with some pecularities. Those will be described in detail below.
Guys, we love you. After some five to ten years of a standard being born, you still cannot manage to get one single new notebook on the market, that actually adheres to that standard. There should be a law against that. Nonetheless, what's the fuss about? With the default kernel of Debian stable (so 2.6.8) nothing will really work. You get a lot of ACPI error messages (and I really mean a lot). So, there is the basic idea, taking a kernel out of unstable and using that one. That doesn't work because of the missing network driver, so I'm currently stuck with a vanilla 2.6.11.12.
With that things look much brighter, however, there is no way of getting the fan under control. My current working hypothesis is, it is not ACPI controlled, and then its runnning through. Not an all too annoying noise, nonetheless, my old Sony ran passive cooled most of the time. I concluded that the fan is internal controlled because it speeds up if the processor is really running.
What else? Well, battery current state is not readable, so basically you don't now whats the status of the battery. Last thing, with acpi enabled, your clock will be running twice as fast. This should be already repaired in 2.6.12 series, however, no luck getting the ATI binary driver running with 2.6.12. Right now, I'm working with ACPI=off. Most annoying thing with the double time clock is, that time for key repeat and double click is dependent on that one. So, have fun... Francois has sent me some helpful comments, this behaviour is related to APIC, it seems to be indeed. Disabeling APIC and ACPI leaves the clock running at normal speed, however, disabeling APIC alone gets my notebook to hang on bootup (some IRQ issue). Kernel 2.6.12 has a fine bootparameter, noapictimer. This works exactly the way I want, however, with 2.6.12 you have to abandon the ati driver. So, wait for 2.6.11.X or an ati driver compatible to 2.6.12.
At least some good news. Ati has finally released an installer for their driver and that one works as well as the nVidia counterpart. Just be sure that you don't use "autodetect" for display, when running fglrxconfig. Autodetect will get confused with two external connectors and shut off the internal display. Use "Laptop Panel Device" as primary display when you want an external display connected. Some confusion here: fglrxconfig will not ask you for the display to use, if you select 'single head', therefore no image on your tft. Select 'clone mode', even if you don't want to use it and then select primary internal and secondary crt.
I've not had any luck connecting a beamer when the notebook is already operational, but then, I haven't tried really hard. It works for me, when I connect the beamer before switching on the notebook.
Switching Display brightness with acpi of 2.6.11.12 works with the function keys without any further intervention. Nontheless, the keycodes will be sent to the konsole itself, but the brightness will be adjusted.
The ati chipset (controller, not vga) is not very well supported with 2.6.8, it seems to wor with 2.6.10 onwards, I have tried 2.6.11.12 and there it is running smoothly.
In principle, the network card is running fine with the tg3 driver. However the Debian team has removed it from the unstable 2.6.11 kernel because of legal concerns. No chance to get it running there, so you need to use a vanilla kernel.
Finally, wlan is also working. I used the ndiswrappers. I downloaded version 1.2 (foregoing the 1.1 provided by debian) with the wireless utils provided by debian. Install the ndiswrappers exactly according to the manual provided on the homepage, as a driver use this one. Specifically chose the bcmwl5.inf out of the BG subdirectory. Your adaptor should be recognized immediately, however, as long as it is not switched on, you won't be able to transmit or receive. If you press the wlan key in front, you should get a message to the console (or dmesg) telling you that keycode xxxx is unknown. If xxxx is e055, wlan is now enabled, if it is e056 it is now disabled. I was content to see all accesspoints to be discovered, I didn't play around with encryption or the like.
I asked someone who has some experience with bluetooth and together we got the mouse working with no sweat. Assuming you find a bluetooth expert or some howto (which I could not find), you should be able to get it running quickly - at least with the bluez driver. The mouse will be connected as an usb mouse, so in debian to be found at /dev/input/mice
Touchpad is a completely ordinary ps2 connected synaptics touchpad. When I first compiled my kernel, I somehow forgot to include the right driver and therefore, I got the touchpad not working. After getting the hardware to work (check this with 'cat /dev/psaux' in Debian) install the synaptics xfree driver package and select synaptics as driver in your xorg or xfree config file
back to main(german)